Fabric and Trainer Codebase Structure

[dangthatsright]

Hello! I had a question about the structure of the codebase so I can more easily debug my own code around using fabric vs trainer.

There are two strategies repo
https://github.com/Lightning-AI/pytorch-lightning/tree/76b691d80c6c5203c66365272ce246ac86e418f0/src/lightning/pytorch/strategies
and
https://github.com/Lightning-AI/pytorch-lightning/tree/76b691d80c6c5203c66365272ce246ac86e418f0/src/lightning/fabric/strategies

I'd imagine the trainer strategies should use the underlying fabric strategies but this doesn't seem to be the case. Furthermore, I'd thought they would be functionally equivalent but that doesn't seem to be the case see: https://lightning.ai/forums/t/saving-fsdp-model-with-custom-fsdpstrategy-results-in-typeerror-cannot-pickle-module-object/5765/2

So my questions are why are there two different strategies, what are the differences between the two, and why might one choose to use one or the other?

Thanks!